I am in my 3rd year of PhD. I am doing my PhD on Computational Biology, with a background of purely Computer Science. Since I am under pressure to publish 4 more journal papers, apart from my review, I am always in a hurry to do the same. What I have noticed for this field is that if I predict something, I need to provide biological validation for my result. Why so?
Is biological validation not the work of biologists? How can a computer science student provide biological validation for his/her prediction? Isn't our job just to predict? Why do the journals want biological validation for the results?
I did a prediction job, where I predicted some proteins to do a particular function. I communicated it to the Molecular BioSystems Journal. They rejected it saying I have no biological validation. If biologists are the one with the final say, then what point do we have in working in this domain? Whatever we predict would be questioned.
-
10"How can a computer science student provide biological validation for his/her prediction?": By collaborating with biologists?– Massimo OrtolanoOct 15, 2016 at 6:29
-
2@Rishika That's a problem of your university or PhD program. A journal cannot and should not care about university rules and about whose thesis the paper goes.– Massimo OrtolanoOct 15, 2016 at 6:42
-
4Probably not, as not all theoretical physicists collaborate with experimental ones: maybe you just chose the wrong venue, and it's that particular journal that requires also a joint experimental validation. But if that it's their editorial policy, you can't expect them to change it because you can't collaborate with biologists. But whatever the policy, you should expect that anything you predict will be questioned, sooner or later, because the purpose of models is to represent the reality.– Massimo OrtolanoOct 15, 2016 at 6:53
-
12The typical way of validating a prediction if you cannot generate experimental data yourself is to use existing (published) data from the literature.– Cape CodeOct 15, 2016 at 7:25
-
8You are not a "computer science student". If you are writing for an audience of computational biologists, you are a computational biologist.– JeffEOct 15, 2016 at 15:57
3 Answers
I am always in a hurry to do the same. What I have noticed for this field is that if I predict something, I need to provide biological validation for my result. Why so?
Because this is how one knows if your results are correct and useful. A prediction is nothing if it hasn't been evaluated. The field is littered with predictions that turn out to be nonsense, only hold up under very narrow circumstances, etc.
Is biological validation not the work of biologists?
No more than it would be if a biologist said "I just developed this algorithm, proving it's correct is the work of a computer scientist."
Beyond that, there's no reason for them to do so. "I've made an arbitrary prediction, anyone want to validate it?" is going to be met with a resounding "No" - the biologists have their own work to do.
How can a computer science student provide biological validation for his/her prediction? Isn't our job just to predict?
No, it's your job to provide useful results. Validated predictions are useful. Just predictions are not.
As people have noted, there are a number of ways to validate a prediction. Existing data, or even simulated data based on known biological patterns, might be sufficient. If no such data exists, it's time to find a collaborator.
Why do the journals want biological validation for the results?
Because anything else is just speculation.
I did a prediction job, where I predicted some proteins to do a particular function. I communicated it to the Molecular BioSystems Journal. They rejected it saying I have no biological validation. If biologists are the one with the final say, then what point do we have in working in this domain? Whatever we predict would be questioned.
Do you know that it was rejected by a biologist? There's a number of computer scientists I know who would have rejected such a paper for having no biological validation. Beyond that, your "point" in working in the field is to generate those predictions and evaluate their correctness. The latter part is also an aspect of good computational biology.
Biological validation doesn't necessarily mean that you need to do experiments to verify the computational results. There are many computational papers published without any accompanying experimental results (of course it would be fantastic if you can, either by yourself or in collaboration with experimentalists.) The real question here is whether you provide a biological context in which to put and assess your work. No matter whether your work is experimental or quantitative, you need to demonstrate that you understand the previous work done on the biological system that you are trying to study: what has been discovered, what are the interesting questions, how do your results build on/confirm/disprove previous work etc. You need to demonstrate that your work is relevant to the biologists working in the field in the sense that it attempts to address the relevant biological questions (or asks a new question that despite its importance has never been considered) and provide unique insight that is difficult if not impossible to obtain from experiments. Biologists are not interested in theory/computation for its own sake, and the failure to connect such work to the experimental reality is one of the biggest stumbling block for people with a "hard science" background working in biology.
Always look at the literature,
If you are doing thermodynamics work, the biological validation will probably come from isothermal titration calorimetry studies. One can get entropy, enthalpy, and Gibbs Free Energy from that alone.
If you are doing protein folding work, the validation will probably come from some form of either fast protein liquid chromatography (FPLC) or High Performance Liquid Chromatography (HPLC) with the main difference being that FPLC appears to be more often used for analysis, purification, and extraction from a complicated matrix, while HPLC is for analysis. The critical difference of either case is the choice of column which is for extraction of hyrophilic/hydrophobic proteins, or specific proteins which are usually retrieved by affinity chromatography binding antigens or antibodies to a column. Another technique is to clone a polyhistadine tag into the a protein from the extracted DNA via polymerase chain reaction (PCR) to exponentially increase the original concentration of DNA. Critical segments must then be spliced with endonucleases and transcribed/translated into proteins. These proteins will have the polyhistidine tag which will bind to nickle in the column and everything else washes out.
If you are doing ensemble work of the functions of whole protein families, then please pick a family of proteins that you would like to study such as the heat shock proteins (HSPs), receptor tyrosinase proteins, histone proteins, efflux pumps, etc. The biological validation for this informatics work using comes from microarray experiments, running lots of proteins on an electrophoresis gel to determine their size, or even mass spectrometry which tells a researcher the those of a protein.
If your research does not match up with the literature, it does not necessarily mean that you are wrong. For example, in the nightshade plant, antifreeze proteins were only found in the plant in the months of November and December, but not in September and October, if I remember correctly. That means that temperature drop induced the formation of the antifreeze proteins only when it was really cold. After the cold season, the proteins are probably just destroyed completely.
Here is a little information about how temperature influences the existence of antifreeze proteins in the plant Nightshade:
http://link.springer.com/article/10.1023/A:1014062714786
The whole protein network of a plant can change in response to chemical stress, heat stress, species differences in a family of plants, age, species differences, or competitive parisitosis. I expect the same to be true for animals, where especially solution is very important. Is the solution fully acqueous, with phosphate buffered saline, blood serum, cerebrospinal fluid, urine, etc?
If you want to ask scientists questions about the matter, check out ResearchGate. It has high marks from lots prominent reporting groups.
-
In addition, you can do more than just predictive biological work. You can create computational models which align with existing biological work.– xyz123Oct 15, 2016 at 23:57
-
Also, check out Bioillumensence Resonance Energy Transfer (BRET). It can experimentally be used to determine if two porteins come close enough to interact by observing a color change where about two flourescent proteins are tagged to a protein which is cloned together in a living cell. Gersting, Søren W., Amelie S. Lotz-Havla, and Ania C. Muntau. "Bioluminescence Resonance Energy Transfer: An Emerging Tool for the Detection of Protein–Protein Interaction in Living Cells." Functional Genomics: Methods and Protocols (2012): 253-263.– xyz123Oct 16, 2016 at 0:07
-
-
It totally does. I explained experimental techniques which validate computational approaches. The questioner is obviously overwhlemed with the task of understanding how his work is related to biological techniques. So I gave him a summary of methods, something that you did not do.– xyz123Oct 16, 2016 at 1:15
-
I think this answer goes too far into the details of specific experiments, etc. If they aren't working specifically on these questions, these details aren't much more useful than just saying "Look at the literature" alone.– AJKOct 16, 2016 at 2:35